Image Reconstruction with DOI Rebinning and PSF Modeling for the Quad-Head PET

For positron emission tomography (PET), the quad-head geometry achieves higher system sensitivity, however severe parallax error also deteriorates the image quality. Using depth decoding detectors and considering a more accurate system response matrix (SRM) in iterative reconstruction are two approaches to correct parallax error. In this work, we implemented an image reconstruction method for the quad-head PET with dual-ended readout depth decoding detectors. Firstly, each coincidence event was assigned to multiple rebinned lines of response (LOR) on a compact sinogram defined by a virtual ring located inside the real detection ring. In addition, we applied a spatially invariant point spread function (PSF) model. The performance of the proposed reconstruction method was evaluated by using Monte Carlo (MC) simulation data, including a uniform circular phantom and a Derenzo phantom. Experimental results show that this method provided lower noise and higher peak-to-valley ratios.